2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of Libav.
6 * Libav is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * Libav is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
27 #include "libavutil/avutil.h"
28 #include "libavutil/bswap.h"
29 #include "libavutil/cpu.h"
30 #include "libavutil/intreadwrite.h"
31 #include "libavutil/mathematics.h"
32 #include "libavutil/pixdesc.h"
36 #include "swscale_internal.h"
38 DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={
39 { 1, 3, 1, 3, 1, 3, 1, 3, },
40 { 2, 0, 2, 0, 2, 0, 2, 0, },
43 DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_8)[2][8]={
44 { 6, 2, 6, 2, 6, 2, 6, 2, },
45 { 0, 4, 0, 4, 0, 4, 0, 4, },
48 DECLARE_ALIGNED(8, const uint8_t, dither_4x4_16)[4][8]={
49 { 8, 4, 11, 7, 8, 4, 11, 7, },
50 { 2, 14, 1, 13, 2, 14, 1, 13, },
51 { 10, 6, 9, 5, 10, 6, 9, 5, },
52 { 0, 12, 3, 15, 0, 12, 3, 15, },
55 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_32)[8][8]={
56 { 17, 9, 23, 15, 16, 8, 22, 14, },
57 { 5, 29, 3, 27, 4, 28, 2, 26, },
58 { 21, 13, 19, 11, 20, 12, 18, 10, },
59 { 0, 24, 6, 30, 1, 25, 7, 31, },
60 { 16, 8, 22, 14, 17, 9, 23, 15, },
61 { 4, 28, 2, 26, 5, 29, 3, 27, },
62 { 20, 12, 18, 10, 21, 13, 19, 11, },
63 { 1, 25, 7, 31, 0, 24, 6, 30, },
66 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_73)[8][8]={
67 { 0, 55, 14, 68, 3, 58, 17, 72, },
68 { 37, 18, 50, 32, 40, 22, 54, 35, },
69 { 9, 64, 5, 59, 13, 67, 8, 63, },
70 { 46, 27, 41, 23, 49, 31, 44, 26, },
71 { 2, 57, 16, 71, 1, 56, 15, 70, },
72 { 39, 21, 52, 34, 38, 19, 51, 33, },
73 { 11, 66, 7, 62, 10, 65, 6, 60, },
74 { 48, 30, 43, 25, 47, 29, 42, 24, },
78 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
79 {117, 62, 158, 103, 113, 58, 155, 100, },
80 { 34, 199, 21, 186, 31, 196, 17, 182, },
81 {144, 89, 131, 76, 141, 86, 127, 72, },
82 { 0, 165, 41, 206, 10, 175, 52, 217, },
83 {110, 55, 151, 96, 120, 65, 162, 107, },
84 { 28, 193, 14, 179, 38, 203, 24, 189, },
85 {138, 83, 124, 69, 148, 93, 134, 79, },
86 { 7, 172, 48, 213, 3, 168, 45, 210, },
89 // tries to correct a gamma of 1.5
90 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
91 { 0, 143, 18, 200, 2, 156, 25, 215, },
92 { 78, 28, 125, 64, 89, 36, 138, 74, },
93 { 10, 180, 3, 161, 16, 195, 8, 175, },
94 {109, 51, 93, 38, 121, 60, 105, 47, },
95 { 1, 152, 23, 210, 0, 147, 20, 205, },
96 { 85, 33, 134, 71, 81, 30, 130, 67, },
97 { 14, 190, 6, 171, 12, 185, 5, 166, },
98 {117, 57, 101, 44, 113, 54, 97, 41, },
101 // tries to correct a gamma of 2.0
102 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
103 { 0, 124, 8, 193, 0, 140, 12, 213, },
104 { 55, 14, 104, 42, 66, 19, 119, 52, },
105 { 3, 168, 1, 145, 6, 187, 3, 162, },
106 { 86, 31, 70, 21, 99, 39, 82, 28, },
107 { 0, 134, 11, 206, 0, 129, 9, 200, },
108 { 62, 17, 114, 48, 58, 16, 109, 45, },
109 { 5, 181, 2, 157, 4, 175, 1, 151, },
110 { 95, 36, 78, 26, 90, 34, 74, 24, },
113 // tries to correct a gamma of 2.5
114 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
115 { 0, 107, 3, 187, 0, 125, 6, 212, },
116 { 39, 7, 86, 28, 49, 11, 102, 36, },
117 { 1, 158, 0, 131, 3, 180, 1, 151, },
118 { 68, 19, 52, 12, 81, 25, 64, 17, },
119 { 0, 119, 5, 203, 0, 113, 4, 195, },
120 { 45, 9, 96, 33, 42, 8, 91, 30, },
121 { 2, 172, 1, 144, 2, 165, 0, 137, },
122 { 77, 23, 60, 15, 72, 21, 56, 14, },
126 #define output_pixel(pos, val, bias, signedness) \
128 AV_WB16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
130 AV_WL16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
133 static av_always_inline void
134 yuv2plane1_16_c_template(const int32_t *src, uint16_t *dest, int dstW,
135 int big_endian, int output_bits)
138 int shift = 19 - output_bits;
140 for (i = 0; i < dstW; i++) {
141 int val = src[i] + (1 << (shift - 1));
142 output_pixel(&dest[i], val, 0, uint);
146 static av_always_inline void
147 yuv2planeX_16_c_template(const int16_t *filter, int filterSize,
148 const int32_t **src, uint16_t *dest, int dstW,
149 int big_endian, int output_bits)
152 int shift = 15 + 16 - output_bits;
154 for (i = 0; i < dstW; i++) {
155 int val = 1 << (30-output_bits);
158 /* range of val is [0,0x7FFFFFFF], so 31 bits, but with lanczos/spline
159 * filters (or anything with negative coeffs, the range can be slightly
160 * wider in both directions. To account for this overflow, we subtract
161 * a constant so it always fits in the signed range (assuming a
162 * reasonable filterSize), and re-add that at the end. */
164 for (j = 0; j < filterSize; j++)
165 val += src[j][i] * filter[j];
167 output_pixel(&dest[i], val, 0x8000, int);
173 #define output_pixel(pos, val) \
175 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
177 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
180 static av_always_inline void
181 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
182 int big_endian, int output_bits)
185 int shift = 15 - output_bits;
187 for (i = 0; i < dstW; i++) {
188 int val = src[i] + (1 << (shift - 1));
189 output_pixel(&dest[i], val);
193 static av_always_inline void
194 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
195 const int16_t **src, uint16_t *dest, int dstW,
196 int big_endian, int output_bits)
199 int shift = 11 + 16 - output_bits;
201 for (i = 0; i < dstW; i++) {
202 int val = 1 << (26-output_bits);
205 for (j = 0; j < filterSize; j++)
206 val += src[j][i] * filter[j];
208 output_pixel(&dest[i], val);
214 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
215 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
216 uint8_t *dest, int dstW, \
217 const uint8_t *dither, int offset)\
219 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
220 (uint16_t *) dest, dstW, is_be, bits); \
222 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
223 const int16_t **src, uint8_t *dest, int dstW, \
224 const uint8_t *dither, int offset)\
226 yuv2planeX_## template_size ## _c_template(filter, \
227 filterSize, (const typeX_t **) src, \
228 (uint16_t *) dest, dstW, is_be, bits); \
230 yuv2NBPS( 9, BE, 1, 10, int16_t)
231 yuv2NBPS( 9, LE, 0, 10, int16_t)
232 yuv2NBPS(10, BE, 1, 10, int16_t)
233 yuv2NBPS(10, LE, 0, 10, int16_t)
234 yuv2NBPS(16, BE, 1, 16, int32_t)
235 yuv2NBPS(16, LE, 0, 16, int32_t)
237 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
238 const int16_t **src, uint8_t *dest, int dstW,
239 const uint8_t *dither, int offset)
242 for (i=0; i<dstW; i++) {
243 int val = dither[(i + offset) & 7] << 12;
245 for (j=0; j<filterSize; j++)
246 val += src[j][i] * filter[j];
248 dest[i]= av_clip_uint8(val>>19);
252 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
253 const uint8_t *dither, int offset)
256 for (i=0; i<dstW; i++) {
257 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
258 dest[i]= av_clip_uint8(val);
262 static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
263 const int16_t **chrUSrc, const int16_t **chrVSrc,
264 uint8_t *dest, int chrDstW)
266 enum PixelFormat dstFormat = c->dstFormat;
267 const uint8_t *chrDither = c->chrDither8;
270 if (dstFormat == PIX_FMT_NV12)
271 for (i=0; i<chrDstW; i++) {
272 int u = chrDither[i & 7] << 12;
273 int v = chrDither[(i + 3) & 7] << 12;
275 for (j=0; j<chrFilterSize; j++) {
276 u += chrUSrc[j][i] * chrFilter[j];
277 v += chrVSrc[j][i] * chrFilter[j];
280 dest[2*i]= av_clip_uint8(u>>19);
281 dest[2*i+1]= av_clip_uint8(v>>19);
284 for (i=0; i<chrDstW; i++) {
285 int u = chrDither[i & 7] << 12;
286 int v = chrDither[(i + 3) & 7] << 12;
288 for (j=0; j<chrFilterSize; j++) {
289 u += chrUSrc[j][i] * chrFilter[j];
290 v += chrVSrc[j][i] * chrFilter[j];
293 dest[2*i]= av_clip_uint8(v>>19);
294 dest[2*i+1]= av_clip_uint8(u>>19);
298 #define accumulate_bit(acc, val) \
300 acc |= (val) >= (128 + 110)
301 #define output_pixel(pos, acc) \
302 if (target == PIX_FMT_MONOBLACK) { \
308 static av_always_inline void
309 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
310 const int16_t **lumSrc, int lumFilterSize,
311 const int16_t *chrFilter, const int16_t **chrUSrc,
312 const int16_t **chrVSrc, int chrFilterSize,
313 const int16_t **alpSrc, uint8_t *dest, int dstW,
314 int y, enum PixelFormat target)
316 const uint8_t * const d128=dither_8x8_220[y&7];
320 for (i = 0; i < dstW; i += 2) {
325 for (j = 0; j < lumFilterSize; j++) {
326 Y1 += lumSrc[j][i] * lumFilter[j];
327 Y2 += lumSrc[j][i+1] * lumFilter[j];
331 if ((Y1 | Y2) & 0x100) {
332 Y1 = av_clip_uint8(Y1);
333 Y2 = av_clip_uint8(Y2);
335 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
336 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
338 output_pixel(*dest++, acc);
343 output_pixel(*dest, acc);
347 static av_always_inline void
348 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
349 const int16_t *ubuf[2], const int16_t *vbuf[2],
350 const int16_t *abuf[2], uint8_t *dest, int dstW,
351 int yalpha, int uvalpha, int y,
352 enum PixelFormat target)
354 const int16_t *buf0 = buf[0], *buf1 = buf[1];
355 const uint8_t * const d128 = dither_8x8_220[y & 7];
356 int yalpha1 = 4096 - yalpha;
359 for (i = 0; i < dstW; i += 8) {
362 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
363 accumulate_bit(acc, Y + d128[0]);
364 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
365 accumulate_bit(acc, Y + d128[1]);
366 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
367 accumulate_bit(acc, Y + d128[2]);
368 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
369 accumulate_bit(acc, Y + d128[3]);
370 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
371 accumulate_bit(acc, Y + d128[4]);
372 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
373 accumulate_bit(acc, Y + d128[5]);
374 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
375 accumulate_bit(acc, Y + d128[6]);
376 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
377 accumulate_bit(acc, Y + d128[7]);
379 output_pixel(*dest++, acc);
383 static av_always_inline void
384 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
385 const int16_t *ubuf[2], const int16_t *vbuf[2],
386 const int16_t *abuf0, uint8_t *dest, int dstW,
387 int uvalpha, int y, enum PixelFormat target)
389 const uint8_t * const d128 = dither_8x8_220[y & 7];
392 for (i = 0; i < dstW; i += 8) {
395 accumulate_bit(acc, (buf0[i + 0] >> 7) + d128[0]);
396 accumulate_bit(acc, (buf0[i + 1] >> 7) + d128[1]);
397 accumulate_bit(acc, (buf0[i + 2] >> 7) + d128[2]);
398 accumulate_bit(acc, (buf0[i + 3] >> 7) + d128[3]);
399 accumulate_bit(acc, (buf0[i + 4] >> 7) + d128[4]);
400 accumulate_bit(acc, (buf0[i + 5] >> 7) + d128[5]);
401 accumulate_bit(acc, (buf0[i + 6] >> 7) + d128[6]);
402 accumulate_bit(acc, (buf0[i + 7] >> 7) + d128[7]);
404 output_pixel(*dest++, acc);
409 #undef accumulate_bit
411 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
412 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
413 const int16_t **lumSrc, int lumFilterSize, \
414 const int16_t *chrFilter, const int16_t **chrUSrc, \
415 const int16_t **chrVSrc, int chrFilterSize, \
416 const int16_t **alpSrc, uint8_t *dest, int dstW, \
419 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
420 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
421 alpSrc, dest, dstW, y, fmt); \
424 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
425 const int16_t *ubuf[2], const int16_t *vbuf[2], \
426 const int16_t *abuf[2], uint8_t *dest, int dstW, \
427 int yalpha, int uvalpha, int y) \
429 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
430 dest, dstW, yalpha, uvalpha, y, fmt); \
433 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
434 const int16_t *ubuf[2], const int16_t *vbuf[2], \
435 const int16_t *abuf0, uint8_t *dest, int dstW, \
436 int uvalpha, int y) \
438 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
439 abuf0, dest, dstW, uvalpha, \
443 YUV2PACKEDWRAPPER(yuv2mono,, white, PIX_FMT_MONOWHITE)
444 YUV2PACKEDWRAPPER(yuv2mono,, black, PIX_FMT_MONOBLACK)
446 #define output_pixels(pos, Y1, U, Y2, V) \
447 if (target == PIX_FMT_YUYV422) { \
448 dest[pos + 0] = Y1; \
450 dest[pos + 2] = Y2; \
454 dest[pos + 1] = Y1; \
456 dest[pos + 3] = Y2; \
459 static av_always_inline void
460 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
461 const int16_t **lumSrc, int lumFilterSize,
462 const int16_t *chrFilter, const int16_t **chrUSrc,
463 const int16_t **chrVSrc, int chrFilterSize,
464 const int16_t **alpSrc, uint8_t *dest, int dstW,
465 int y, enum PixelFormat target)
469 for (i = 0; i < ((dstW + 1) >> 1); i++) {
476 for (j = 0; j < lumFilterSize; j++) {
477 Y1 += lumSrc[j][i * 2] * lumFilter[j];
478 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
480 for (j = 0; j < chrFilterSize; j++) {
481 U += chrUSrc[j][i] * chrFilter[j];
482 V += chrVSrc[j][i] * chrFilter[j];
488 if ((Y1 | Y2 | U | V) & 0x100) {
489 Y1 = av_clip_uint8(Y1);
490 Y2 = av_clip_uint8(Y2);
491 U = av_clip_uint8(U);
492 V = av_clip_uint8(V);
494 output_pixels(4*i, Y1, U, Y2, V);
498 static av_always_inline void
499 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
500 const int16_t *ubuf[2], const int16_t *vbuf[2],
501 const int16_t *abuf[2], uint8_t *dest, int dstW,
502 int yalpha, int uvalpha, int y,
503 enum PixelFormat target)
505 const int16_t *buf0 = buf[0], *buf1 = buf[1],
506 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
507 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
508 int yalpha1 = 4096 - yalpha;
509 int uvalpha1 = 4096 - uvalpha;
512 for (i = 0; i < ((dstW + 1) >> 1); i++) {
513 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
514 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
515 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
516 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
518 Y1 = av_clip_uint8(Y1);
519 Y2 = av_clip_uint8(Y2);
520 U = av_clip_uint8(U);
521 V = av_clip_uint8(V);
523 output_pixels(i * 4, Y1, U, Y2, V);
527 static av_always_inline void
528 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
529 const int16_t *ubuf[2], const int16_t *vbuf[2],
530 const int16_t *abuf0, uint8_t *dest, int dstW,
531 int uvalpha, int y, enum PixelFormat target)
533 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
536 if (uvalpha < 2048) {
537 for (i = 0; i < ((dstW + 1) >> 1); i++) {
538 int Y1 = buf0[i * 2] >> 7;
539 int Y2 = buf0[i * 2 + 1] >> 7;
540 int U = ubuf0[i] >> 7;
541 int V = vbuf0[i] >> 7;
543 Y1 = av_clip_uint8(Y1);
544 Y2 = av_clip_uint8(Y2);
545 U = av_clip_uint8(U);
546 V = av_clip_uint8(V);
548 output_pixels(i * 4, Y1, U, Y2, V);
551 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
552 for (i = 0; i < ((dstW + 1) >> 1); i++) {
553 int Y1 = buf0[i * 2] >> 7;
554 int Y2 = buf0[i * 2 + 1] >> 7;
555 int U = (ubuf0[i] + ubuf1[i]) >> 8;
556 int V = (vbuf0[i] + vbuf1[i]) >> 8;
558 Y1 = av_clip_uint8(Y1);
559 Y2 = av_clip_uint8(Y2);
560 U = av_clip_uint8(U);
561 V = av_clip_uint8(V);
563 output_pixels(i * 4, Y1, U, Y2, V);
570 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, PIX_FMT_YUYV422)
571 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, PIX_FMT_UYVY422)
573 #define R_B ((target == PIX_FMT_RGB48LE || target == PIX_FMT_RGB48BE) ? R : B)
574 #define B_R ((target == PIX_FMT_RGB48LE || target == PIX_FMT_RGB48BE) ? B : R)
575 #define output_pixel(pos, val) \
576 if (isBE(target)) { \
582 static av_always_inline void
583 yuv2rgb48_X_c_template(SwsContext *c, const int16_t *lumFilter,
584 const int32_t **lumSrc, int lumFilterSize,
585 const int16_t *chrFilter, const int32_t **chrUSrc,
586 const int32_t **chrVSrc, int chrFilterSize,
587 const int32_t **alpSrc, uint16_t *dest, int dstW,
588 int y, enum PixelFormat target)
592 for (i = 0; i < ((dstW + 1) >> 1); i++) {
594 int Y1 = -0x40000000;
595 int Y2 = -0x40000000;
596 int U = -128 << 23; // 19
600 for (j = 0; j < lumFilterSize; j++) {
601 Y1 += lumSrc[j][i * 2] * lumFilter[j];
602 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
604 for (j = 0; j < chrFilterSize; j++) {
605 U += chrUSrc[j][i] * chrFilter[j];
606 V += chrVSrc[j][i] * chrFilter[j];
609 // 8bit: 12+15=27; 16-bit: 12+19=31
617 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
618 Y1 -= c->yuv2rgb_y_offset;
619 Y2 -= c->yuv2rgb_y_offset;
620 Y1 *= c->yuv2rgb_y_coeff;
621 Y2 *= c->yuv2rgb_y_coeff;
624 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
626 R = V * c->yuv2rgb_v2r_coeff;
627 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
628 B = U * c->yuv2rgb_u2b_coeff;
630 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
631 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
632 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
633 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
634 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
635 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
636 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
641 static av_always_inline void
642 yuv2rgb48_2_c_template(SwsContext *c, const int32_t *buf[2],
643 const int32_t *ubuf[2], const int32_t *vbuf[2],
644 const int32_t *abuf[2], uint16_t *dest, int dstW,
645 int yalpha, int uvalpha, int y,
646 enum PixelFormat target)
648 const int32_t *buf0 = buf[0], *buf1 = buf[1],
649 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
650 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
651 int yalpha1 = 4096 - yalpha;
652 int uvalpha1 = 4096 - uvalpha;
655 for (i = 0; i < ((dstW + 1) >> 1); i++) {
656 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
657 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
658 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha + (-128 << 23)) >> 14;
659 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha + (-128 << 23)) >> 14;
662 Y1 -= c->yuv2rgb_y_offset;
663 Y2 -= c->yuv2rgb_y_offset;
664 Y1 *= c->yuv2rgb_y_coeff;
665 Y2 *= c->yuv2rgb_y_coeff;
669 R = V * c->yuv2rgb_v2r_coeff;
670 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
671 B = U * c->yuv2rgb_u2b_coeff;
673 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
674 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
675 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
676 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
677 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
678 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
683 static av_always_inline void
684 yuv2rgb48_1_c_template(SwsContext *c, const int32_t *buf0,
685 const int32_t *ubuf[2], const int32_t *vbuf[2],
686 const int32_t *abuf0, uint16_t *dest, int dstW,
687 int uvalpha, int y, enum PixelFormat target)
689 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
692 if (uvalpha < 2048) {
693 for (i = 0; i < ((dstW + 1) >> 1); i++) {
694 int Y1 = (buf0[i * 2] ) >> 2;
695 int Y2 = (buf0[i * 2 + 1]) >> 2;
696 int U = (ubuf0[i] + (-128 << 11)) >> 2;
697 int V = (vbuf0[i] + (-128 << 11)) >> 2;
700 Y1 -= c->yuv2rgb_y_offset;
701 Y2 -= c->yuv2rgb_y_offset;
702 Y1 *= c->yuv2rgb_y_coeff;
703 Y2 *= c->yuv2rgb_y_coeff;
707 R = V * c->yuv2rgb_v2r_coeff;
708 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
709 B = U * c->yuv2rgb_u2b_coeff;
711 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
712 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
713 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
714 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
715 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
716 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
720 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
721 for (i = 0; i < ((dstW + 1) >> 1); i++) {
722 int Y1 = (buf0[i * 2] ) >> 2;
723 int Y2 = (buf0[i * 2 + 1]) >> 2;
724 int U = (ubuf0[i] + ubuf1[i] + (-128 << 12)) >> 3;
725 int V = (vbuf0[i] + vbuf1[i] + (-128 << 12)) >> 3;
728 Y1 -= c->yuv2rgb_y_offset;
729 Y2 -= c->yuv2rgb_y_offset;
730 Y1 *= c->yuv2rgb_y_coeff;
731 Y2 *= c->yuv2rgb_y_coeff;
735 R = V * c->yuv2rgb_v2r_coeff;
736 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
737 B = U * c->yuv2rgb_u2b_coeff;
739 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
740 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
741 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
742 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
743 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
744 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
754 #define YUV2PACKED16WRAPPER(name, base, ext, fmt) \
755 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
756 const int16_t **_lumSrc, int lumFilterSize, \
757 const int16_t *chrFilter, const int16_t **_chrUSrc, \
758 const int16_t **_chrVSrc, int chrFilterSize, \
759 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
762 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
763 **chrUSrc = (const int32_t **) _chrUSrc, \
764 **chrVSrc = (const int32_t **) _chrVSrc, \
765 **alpSrc = (const int32_t **) _alpSrc; \
766 uint16_t *dest = (uint16_t *) _dest; \
767 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
768 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
769 alpSrc, dest, dstW, y, fmt); \
772 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
773 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
774 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
775 int yalpha, int uvalpha, int y) \
777 const int32_t **buf = (const int32_t **) _buf, \
778 **ubuf = (const int32_t **) _ubuf, \
779 **vbuf = (const int32_t **) _vbuf, \
780 **abuf = (const int32_t **) _abuf; \
781 uint16_t *dest = (uint16_t *) _dest; \
782 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
783 dest, dstW, yalpha, uvalpha, y, fmt); \
786 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
787 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
788 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
789 int uvalpha, int y) \
791 const int32_t *buf0 = (const int32_t *) _buf0, \
792 **ubuf = (const int32_t **) _ubuf, \
793 **vbuf = (const int32_t **) _vbuf, \
794 *abuf0 = (const int32_t *) _abuf0; \
795 uint16_t *dest = (uint16_t *) _dest; \
796 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
797 dstW, uvalpha, y, fmt); \
800 YUV2PACKED16WRAPPER(yuv2, rgb48, rgb48be, PIX_FMT_RGB48BE)
801 YUV2PACKED16WRAPPER(yuv2, rgb48, rgb48le, PIX_FMT_RGB48LE)
802 YUV2PACKED16WRAPPER(yuv2, rgb48, bgr48be, PIX_FMT_BGR48BE)
803 YUV2PACKED16WRAPPER(yuv2, rgb48, bgr48le, PIX_FMT_BGR48LE)
806 * Write out 2 RGB pixels in the target pixel format. This function takes a
807 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
808 * things like endianness conversion and shifting. The caller takes care of
809 * setting the correct offset in these tables from the chroma (U/V) values.
810 * This function then uses the luminance (Y1/Y2) values to write out the
811 * correct RGB values into the destination buffer.
813 static av_always_inline void
814 yuv2rgb_write(uint8_t *_dest, int i, unsigned Y1, unsigned Y2,
815 unsigned A1, unsigned A2,
816 const void *_r, const void *_g, const void *_b, int y,
817 enum PixelFormat target, int hasAlpha)
819 if (target == PIX_FMT_ARGB || target == PIX_FMT_RGBA ||
820 target == PIX_FMT_ABGR || target == PIX_FMT_BGRA) {
821 uint32_t *dest = (uint32_t *) _dest;
822 const uint32_t *r = (const uint32_t *) _r;
823 const uint32_t *g = (const uint32_t *) _g;
824 const uint32_t *b = (const uint32_t *) _b;
827 int sh = hasAlpha ? ((target == PIX_FMT_RGB32_1 || target == PIX_FMT_BGR32_1) ? 0 : 24) : 0;
829 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
830 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
833 int sh = (target == PIX_FMT_RGB32_1 || target == PIX_FMT_BGR32_1) ? 0 : 24;
835 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
836 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
838 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
839 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
842 } else if (target == PIX_FMT_RGB24 || target == PIX_FMT_BGR24) {
843 uint8_t *dest = (uint8_t *) _dest;
844 const uint8_t *r = (const uint8_t *) _r;
845 const uint8_t *g = (const uint8_t *) _g;
846 const uint8_t *b = (const uint8_t *) _b;
848 #define r_b ((target == PIX_FMT_RGB24) ? r : b)
849 #define b_r ((target == PIX_FMT_RGB24) ? b : r)
850 dest[i * 6 + 0] = r_b[Y1];
851 dest[i * 6 + 1] = g[Y1];
852 dest[i * 6 + 2] = b_r[Y1];
853 dest[i * 6 + 3] = r_b[Y2];
854 dest[i * 6 + 4] = g[Y2];
855 dest[i * 6 + 5] = b_r[Y2];
858 } else if (target == PIX_FMT_RGB565 || target == PIX_FMT_BGR565 ||
859 target == PIX_FMT_RGB555 || target == PIX_FMT_BGR555 ||
860 target == PIX_FMT_RGB444 || target == PIX_FMT_BGR444) {
861 uint16_t *dest = (uint16_t *) _dest;
862 const uint16_t *r = (const uint16_t *) _r;
863 const uint16_t *g = (const uint16_t *) _g;
864 const uint16_t *b = (const uint16_t *) _b;
865 int dr1, dg1, db1, dr2, dg2, db2;
867 if (target == PIX_FMT_RGB565 || target == PIX_FMT_BGR565) {
868 dr1 = dither_2x2_8[ y & 1 ][0];
869 dg1 = dither_2x2_4[ y & 1 ][0];
870 db1 = dither_2x2_8[(y & 1) ^ 1][0];
871 dr2 = dither_2x2_8[ y & 1 ][1];
872 dg2 = dither_2x2_4[ y & 1 ][1];
873 db2 = dither_2x2_8[(y & 1) ^ 1][1];
874 } else if (target == PIX_FMT_RGB555 || target == PIX_FMT_BGR555) {
875 dr1 = dither_2x2_8[ y & 1 ][0];
876 dg1 = dither_2x2_8[ y & 1 ][1];
877 db1 = dither_2x2_8[(y & 1) ^ 1][0];
878 dr2 = dither_2x2_8[ y & 1 ][1];
879 dg2 = dither_2x2_8[ y & 1 ][0];
880 db2 = dither_2x2_8[(y & 1) ^ 1][1];
882 dr1 = dither_4x4_16[ y & 3 ][0];
883 dg1 = dither_4x4_16[ y & 3 ][1];
884 db1 = dither_4x4_16[(y & 3) ^ 3][0];
885 dr2 = dither_4x4_16[ y & 3 ][1];
886 dg2 = dither_4x4_16[ y & 3 ][0];
887 db2 = dither_4x4_16[(y & 3) ^ 3][1];
890 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
891 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
892 } else /* 8/4-bit */ {
893 uint8_t *dest = (uint8_t *) _dest;
894 const uint8_t *r = (const uint8_t *) _r;
895 const uint8_t *g = (const uint8_t *) _g;
896 const uint8_t *b = (const uint8_t *) _b;
897 int dr1, dg1, db1, dr2, dg2, db2;
899 if (target == PIX_FMT_RGB8 || target == PIX_FMT_BGR8) {
900 const uint8_t * const d64 = dither_8x8_73[y & 7];
901 const uint8_t * const d32 = dither_8x8_32[y & 7];
902 dr1 = dg1 = d32[(i * 2 + 0) & 7];
903 db1 = d64[(i * 2 + 0) & 7];
904 dr2 = dg2 = d32[(i * 2 + 1) & 7];
905 db2 = d64[(i * 2 + 1) & 7];
907 const uint8_t * const d64 = dither_8x8_73 [y & 7];
908 const uint8_t * const d128 = dither_8x8_220[y & 7];
909 dr1 = db1 = d128[(i * 2 + 0) & 7];
910 dg1 = d64[(i * 2 + 0) & 7];
911 dr2 = db2 = d128[(i * 2 + 1) & 7];
912 dg2 = d64[(i * 2 + 1) & 7];
915 if (target == PIX_FMT_RGB4 || target == PIX_FMT_BGR4) {
916 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
917 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
919 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
920 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
925 static av_always_inline void
926 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
927 const int16_t **lumSrc, int lumFilterSize,
928 const int16_t *chrFilter, const int16_t **chrUSrc,
929 const int16_t **chrVSrc, int chrFilterSize,
930 const int16_t **alpSrc, uint8_t *dest, int dstW,
931 int y, enum PixelFormat target, int hasAlpha)
935 for (i = 0; i < ((dstW + 1) >> 1); i++) {
941 const void *r, *g, *b;
943 for (j = 0; j < lumFilterSize; j++) {
944 Y1 += lumSrc[j][i * 2] * lumFilter[j];
945 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
947 for (j = 0; j < chrFilterSize; j++) {
948 U += chrUSrc[j][i] * chrFilter[j];
949 V += chrVSrc[j][i] * chrFilter[j];
955 if ((Y1 | Y2 | U | V) & 0x100) {
956 Y1 = av_clip_uint8(Y1);
957 Y2 = av_clip_uint8(Y2);
958 U = av_clip_uint8(U);
959 V = av_clip_uint8(V);
964 for (j = 0; j < lumFilterSize; j++) {
965 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
966 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
970 if ((A1 | A2) & 0x100) {
971 A1 = av_clip_uint8(A1);
972 A2 = av_clip_uint8(A2);
976 /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/
978 g = (c->table_gU[U] + c->table_gV[V]);
981 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
982 r, g, b, y, target, hasAlpha);
986 static av_always_inline void
987 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
988 const int16_t *ubuf[2], const int16_t *vbuf[2],
989 const int16_t *abuf[2], uint8_t *dest, int dstW,
990 int yalpha, int uvalpha, int y,
991 enum PixelFormat target, int hasAlpha)
993 const int16_t *buf0 = buf[0], *buf1 = buf[1],
994 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
995 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
996 *abuf0 = hasAlpha ? abuf[0] : NULL,
997 *abuf1 = hasAlpha ? abuf[1] : NULL;
998 int yalpha1 = 4096 - yalpha;
999 int uvalpha1 = 4096 - uvalpha;
1002 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1003 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1004 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1005 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1006 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1008 const void *r, *g, *b;
1010 Y1 = av_clip_uint8(Y1);
1011 Y2 = av_clip_uint8(Y2);
1012 U = av_clip_uint8(U);
1013 V = av_clip_uint8(V);
1016 g = (c->table_gU[U] + c->table_gV[V]);
1020 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1021 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1022 A1 = av_clip_uint8(A1);
1023 A2 = av_clip_uint8(A2);
1026 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1027 r, g, b, y, target, hasAlpha);
1031 static av_always_inline void
1032 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1033 const int16_t *ubuf[2], const int16_t *vbuf[2],
1034 const int16_t *abuf0, uint8_t *dest, int dstW,
1035 int uvalpha, int y, enum PixelFormat target,
1038 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1041 if (uvalpha < 2048) {
1042 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1043 int Y1 = buf0[i * 2] >> 7;
1044 int Y2 = buf0[i * 2 + 1] >> 7;
1045 int U = ubuf0[i] >> 7;
1046 int V = vbuf0[i] >> 7;
1048 const void *r, *g, *b;
1050 Y1 = av_clip_uint8(Y1);
1051 Y2 = av_clip_uint8(Y2);
1052 U = av_clip_uint8(U);
1053 V = av_clip_uint8(V);
1056 g = (c->table_gU[U] + c->table_gV[V]);
1060 A1 = abuf0[i * 2 ] >> 7;
1061 A2 = abuf0[i * 2 + 1] >> 7;
1062 A1 = av_clip_uint8(A1);
1063 A2 = av_clip_uint8(A2);
1066 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1067 r, g, b, y, target, hasAlpha);
1070 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1071 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1072 int Y1 = buf0[i * 2] >> 7;
1073 int Y2 = buf0[i * 2 + 1] >> 7;
1074 int U = (ubuf0[i] + ubuf1[i]) >> 8;
1075 int V = (vbuf0[i] + vbuf1[i]) >> 8;
1077 const void *r, *g, *b;
1079 Y1 = av_clip_uint8(Y1);
1080 Y2 = av_clip_uint8(Y2);
1081 U = av_clip_uint8(U);
1082 V = av_clip_uint8(V);
1085 g = (c->table_gU[U] + c->table_gV[V]);
1089 A1 = abuf0[i * 2 ] >> 7;
1090 A2 = abuf0[i * 2 + 1] >> 7;
1091 A1 = av_clip_uint8(A1);
1092 A2 = av_clip_uint8(A2);
1095 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1096 r, g, b, y, target, hasAlpha);
1101 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1102 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1103 const int16_t **lumSrc, int lumFilterSize, \
1104 const int16_t *chrFilter, const int16_t **chrUSrc, \
1105 const int16_t **chrVSrc, int chrFilterSize, \
1106 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1109 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1110 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1111 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1113 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1114 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1115 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1116 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1117 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1118 int yalpha, int uvalpha, int y) \
1120 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1121 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1124 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1125 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1126 const int16_t *abuf0, uint8_t *dest, int dstW, \
1127 int uvalpha, int y) \
1129 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1130 dstW, uvalpha, y, fmt, hasAlpha); \
1134 YUV2RGBWRAPPER(yuv2rgb,, 32_1, PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1135 YUV2RGBWRAPPER(yuv2rgb,, 32, PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1137 #if CONFIG_SWSCALE_ALPHA
1138 YUV2RGBWRAPPER(yuv2rgb,, a32_1, PIX_FMT_RGB32_1, 1)
1139 YUV2RGBWRAPPER(yuv2rgb,, a32, PIX_FMT_RGB32, 1)
1141 YUV2RGBWRAPPER(yuv2rgb,, x32_1, PIX_FMT_RGB32_1, 0)
1142 YUV2RGBWRAPPER(yuv2rgb,, x32, PIX_FMT_RGB32, 0)
1144 YUV2RGBWRAPPER(yuv2, rgb, rgb24, PIX_FMT_RGB24, 0)
1145 YUV2RGBWRAPPER(yuv2, rgb, bgr24, PIX_FMT_BGR24, 0)
1146 YUV2RGBWRAPPER(yuv2rgb,, 16, PIX_FMT_RGB565, 0)
1147 YUV2RGBWRAPPER(yuv2rgb,, 15, PIX_FMT_RGB555, 0)
1148 YUV2RGBWRAPPER(yuv2rgb,, 12, PIX_FMT_RGB444, 0)
1149 YUV2RGBWRAPPER(yuv2rgb,, 8, PIX_FMT_RGB8, 0)
1150 YUV2RGBWRAPPER(yuv2rgb,, 4, PIX_FMT_RGB4, 0)
1151 YUV2RGBWRAPPER(yuv2rgb,, 4b, PIX_FMT_RGB4_BYTE, 0)
1153 static av_always_inline void
1154 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1155 const int16_t **lumSrc, int lumFilterSize,
1156 const int16_t *chrFilter, const int16_t **chrUSrc,
1157 const int16_t **chrVSrc, int chrFilterSize,
1158 const int16_t **alpSrc, uint8_t *dest,
1159 int dstW, int y, enum PixelFormat target, int hasAlpha)
1162 int step = (target == PIX_FMT_RGB24 || target == PIX_FMT_BGR24) ? 3 : 4;
1164 for (i = 0; i < dstW; i++) {
1171 for (j = 0; j < lumFilterSize; j++) {
1172 Y += lumSrc[j][i] * lumFilter[j];
1174 for (j = 0; j < chrFilterSize; j++) {
1175 U += chrUSrc[j][i] * chrFilter[j];
1176 V += chrVSrc[j][i] * chrFilter[j];
1183 for (j = 0; j < lumFilterSize; j++) {
1184 A += alpSrc[j][i] * lumFilter[j];
1188 A = av_clip_uint8(A);
1190 Y -= c->yuv2rgb_y_offset;
1191 Y *= c->yuv2rgb_y_coeff;
1193 R = Y + V*c->yuv2rgb_v2r_coeff;
1194 G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1195 B = Y + U*c->yuv2rgb_u2b_coeff;
1196 if ((R | G | B) & 0xC0000000) {
1197 R = av_clip_uintp2(R, 30);
1198 G = av_clip_uintp2(G, 30);
1199 B = av_clip_uintp2(B, 30);
1204 dest[0] = hasAlpha ? A : 255;
1218 dest[3] = hasAlpha ? A : 255;
1221 dest[0] = hasAlpha ? A : 255;
1236 dest[3] = hasAlpha ? A : 255;
1244 YUV2RGBWRAPPERX(yuv2, rgb_full, bgra32_full, PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1245 YUV2RGBWRAPPERX(yuv2, rgb_full, abgr32_full, PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1246 YUV2RGBWRAPPERX(yuv2, rgb_full, rgba32_full, PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1247 YUV2RGBWRAPPERX(yuv2, rgb_full, argb32_full, PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1249 #if CONFIG_SWSCALE_ALPHA
1250 YUV2RGBWRAPPERX(yuv2, rgb_full, bgra32_full, PIX_FMT_BGRA, 1)
1251 YUV2RGBWRAPPERX(yuv2, rgb_full, abgr32_full, PIX_FMT_ABGR, 1)
1252 YUV2RGBWRAPPERX(yuv2, rgb_full, rgba32_full, PIX_FMT_RGBA, 1)
1253 YUV2RGBWRAPPERX(yuv2, rgb_full, argb32_full, PIX_FMT_ARGB, 1)
1255 YUV2RGBWRAPPERX(yuv2, rgb_full, bgrx32_full, PIX_FMT_BGRA, 0)
1256 YUV2RGBWRAPPERX(yuv2, rgb_full, xbgr32_full, PIX_FMT_ABGR, 0)
1257 YUV2RGBWRAPPERX(yuv2, rgb_full, rgbx32_full, PIX_FMT_RGBA, 0)
1258 YUV2RGBWRAPPERX(yuv2, rgb_full, xrgb32_full, PIX_FMT_ARGB, 0)
1260 YUV2RGBWRAPPERX(yuv2, rgb_full, bgr24_full, PIX_FMT_BGR24, 0)
1261 YUV2RGBWRAPPERX(yuv2, rgb_full, rgb24_full, PIX_FMT_RGB24, 0)
1263 void ff_sws_init_output_funcs(SwsContext *c,
1264 yuv2planar1_fn *yuv2plane1,
1265 yuv2planarX_fn *yuv2planeX,
1266 yuv2interleavedX_fn *yuv2nv12cX,
1267 yuv2packed1_fn *yuv2packed1,
1268 yuv2packed2_fn *yuv2packed2,
1269 yuv2packedX_fn *yuv2packedX)
1271 enum PixelFormat dstFormat = c->dstFormat;
1273 if (is16BPS(dstFormat)) {
1274 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
1275 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
1276 } else if (is9_OR_10BPS(dstFormat)) {
1277 if (av_pix_fmt_descriptors[dstFormat].comp[0].depth_minus1 == 8) {
1278 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
1279 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
1281 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
1282 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
1285 *yuv2plane1 = yuv2plane1_8_c;
1286 *yuv2planeX = yuv2planeX_8_c;
1287 if (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)
1288 *yuv2nv12cX = yuv2nv12cX_c;
1291 if(c->flags & SWS_FULL_CHR_H_INT) {
1292 switch (dstFormat) {
1295 *yuv2packedX = yuv2rgba32_full_X_c;
1297 #if CONFIG_SWSCALE_ALPHA
1299 *yuv2packedX = yuv2rgba32_full_X_c;
1301 #endif /* CONFIG_SWSCALE_ALPHA */
1303 *yuv2packedX = yuv2rgbx32_full_X_c;
1305 #endif /* !CONFIG_SMALL */
1309 *yuv2packedX = yuv2argb32_full_X_c;
1311 #if CONFIG_SWSCALE_ALPHA
1313 *yuv2packedX = yuv2argb32_full_X_c;
1315 #endif /* CONFIG_SWSCALE_ALPHA */
1317 *yuv2packedX = yuv2xrgb32_full_X_c;
1319 #endif /* !CONFIG_SMALL */
1323 *yuv2packedX = yuv2bgra32_full_X_c;
1325 #if CONFIG_SWSCALE_ALPHA
1327 *yuv2packedX = yuv2bgra32_full_X_c;
1329 #endif /* CONFIG_SWSCALE_ALPHA */
1331 *yuv2packedX = yuv2bgrx32_full_X_c;
1333 #endif /* !CONFIG_SMALL */
1337 *yuv2packedX = yuv2abgr32_full_X_c;
1339 #if CONFIG_SWSCALE_ALPHA
1341 *yuv2packedX = yuv2abgr32_full_X_c;
1343 #endif /* CONFIG_SWSCALE_ALPHA */
1345 *yuv2packedX = yuv2xbgr32_full_X_c;
1347 #endif /* !CONFIG_SMALL */
1350 *yuv2packedX = yuv2rgb24_full_X_c;
1353 *yuv2packedX = yuv2bgr24_full_X_c;
1357 switch (dstFormat) {
1358 case PIX_FMT_RGB48LE:
1359 *yuv2packed1 = yuv2rgb48le_1_c;
1360 *yuv2packed2 = yuv2rgb48le_2_c;
1361 *yuv2packedX = yuv2rgb48le_X_c;
1363 case PIX_FMT_RGB48BE:
1364 *yuv2packed1 = yuv2rgb48be_1_c;
1365 *yuv2packed2 = yuv2rgb48be_2_c;
1366 *yuv2packedX = yuv2rgb48be_X_c;
1368 case PIX_FMT_BGR48LE:
1369 *yuv2packed1 = yuv2bgr48le_1_c;
1370 *yuv2packed2 = yuv2bgr48le_2_c;
1371 *yuv2packedX = yuv2bgr48le_X_c;
1373 case PIX_FMT_BGR48BE:
1374 *yuv2packed1 = yuv2bgr48be_1_c;
1375 *yuv2packed2 = yuv2bgr48be_2_c;
1376 *yuv2packedX = yuv2bgr48be_X_c;
1381 *yuv2packed1 = yuv2rgb32_1_c;
1382 *yuv2packed2 = yuv2rgb32_2_c;
1383 *yuv2packedX = yuv2rgb32_X_c;
1385 #if CONFIG_SWSCALE_ALPHA
1387 *yuv2packed1 = yuv2rgba32_1_c;
1388 *yuv2packed2 = yuv2rgba32_2_c;
1389 *yuv2packedX = yuv2rgba32_X_c;
1391 #endif /* CONFIG_SWSCALE_ALPHA */
1393 *yuv2packed1 = yuv2rgbx32_1_c;
1394 *yuv2packed2 = yuv2rgbx32_2_c;
1395 *yuv2packedX = yuv2rgbx32_X_c;
1397 #endif /* !CONFIG_SMALL */
1399 case PIX_FMT_RGB32_1:
1400 case PIX_FMT_BGR32_1:
1402 *yuv2packed1 = yuv2rgb32_1_1_c;
1403 *yuv2packed2 = yuv2rgb32_1_2_c;
1404 *yuv2packedX = yuv2rgb32_1_X_c;
1406 #if CONFIG_SWSCALE_ALPHA
1408 *yuv2packed1 = yuv2rgba32_1_1_c;
1409 *yuv2packed2 = yuv2rgba32_1_2_c;
1410 *yuv2packedX = yuv2rgba32_1_X_c;
1412 #endif /* CONFIG_SWSCALE_ALPHA */
1414 *yuv2packed1 = yuv2rgbx32_1_1_c;
1415 *yuv2packed2 = yuv2rgbx32_1_2_c;
1416 *yuv2packedX = yuv2rgbx32_1_X_c;
1418 #endif /* !CONFIG_SMALL */
1421 *yuv2packed1 = yuv2rgb24_1_c;
1422 *yuv2packed2 = yuv2rgb24_2_c;
1423 *yuv2packedX = yuv2rgb24_X_c;
1426 *yuv2packed1 = yuv2bgr24_1_c;
1427 *yuv2packed2 = yuv2bgr24_2_c;
1428 *yuv2packedX = yuv2bgr24_X_c;
1430 case PIX_FMT_RGB565LE:
1431 case PIX_FMT_RGB565BE:
1432 case PIX_FMT_BGR565LE:
1433 case PIX_FMT_BGR565BE:
1434 *yuv2packed1 = yuv2rgb16_1_c;
1435 *yuv2packed2 = yuv2rgb16_2_c;
1436 *yuv2packedX = yuv2rgb16_X_c;
1438 case PIX_FMT_RGB555LE:
1439 case PIX_FMT_RGB555BE:
1440 case PIX_FMT_BGR555LE:
1441 case PIX_FMT_BGR555BE:
1442 *yuv2packed1 = yuv2rgb15_1_c;
1443 *yuv2packed2 = yuv2rgb15_2_c;
1444 *yuv2packedX = yuv2rgb15_X_c;
1446 case PIX_FMT_RGB444LE:
1447 case PIX_FMT_RGB444BE:
1448 case PIX_FMT_BGR444LE:
1449 case PIX_FMT_BGR444BE:
1450 *yuv2packed1 = yuv2rgb12_1_c;
1451 *yuv2packed2 = yuv2rgb12_2_c;
1452 *yuv2packedX = yuv2rgb12_X_c;
1456 *yuv2packed1 = yuv2rgb8_1_c;
1457 *yuv2packed2 = yuv2rgb8_2_c;
1458 *yuv2packedX = yuv2rgb8_X_c;
1462 *yuv2packed1 = yuv2rgb4_1_c;
1463 *yuv2packed2 = yuv2rgb4_2_c;
1464 *yuv2packedX = yuv2rgb4_X_c;
1466 case PIX_FMT_RGB4_BYTE:
1467 case PIX_FMT_BGR4_BYTE:
1468 *yuv2packed1 = yuv2rgb4b_1_c;
1469 *yuv2packed2 = yuv2rgb4b_2_c;
1470 *yuv2packedX = yuv2rgb4b_X_c;
1474 switch (dstFormat) {
1475 case PIX_FMT_MONOWHITE:
1476 *yuv2packed1 = yuv2monowhite_1_c;
1477 *yuv2packed2 = yuv2monowhite_2_c;
1478 *yuv2packedX = yuv2monowhite_X_c;
1480 case PIX_FMT_MONOBLACK:
1481 *yuv2packed1 = yuv2monoblack_1_c;
1482 *yuv2packed2 = yuv2monoblack_2_c;
1483 *yuv2packedX = yuv2monoblack_X_c;
1485 case PIX_FMT_YUYV422:
1486 *yuv2packed1 = yuv2yuyv422_1_c;
1487 *yuv2packed2 = yuv2yuyv422_2_c;
1488 *yuv2packedX = yuv2yuyv422_X_c;
1490 case PIX_FMT_UYVY422:
1491 *yuv2packed1 = yuv2uyvy422_1_c;
1492 *yuv2packed2 = yuv2uyvy422_2_c;
1493 *yuv2packedX = yuv2uyvy422_X_c;